codeguard-astro
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes user-provided Astro project files and linting configurations which may contain instructions intended to influence the agent.
- Ingestion points: .astro files and ESLint config files including eslint.config.js, .eslintrc.json, .eslintrc.js, .eslintrc.yml, and .pre-commit-config.yaml.
- Capability inventory: Shell execution of npx commands for linting and formatting with read/write access to project files.
- Boundary markers: Results are structured using a specific markdown template ('Astro Codeguard 结果') to delimit tool output.
- Sanitization: The instructions explicitly require the agent to present diffs for human review before applying fixes and obtain confirmation for non-trivial changes.
- [COMMAND_EXECUTION]: The skill uses shell commands to run standard development tools.
- Evidence: It executes 'npx --no-install eslint .' for read-only validation and 'npx eslint . --fix' for applying corrections. These commands are localized to the project environment and are used according to the skill's stated purpose.
Audit Metadata